https://www.metabase.com/blog/Metabase-0.41/index.html

* fix expression dimension failing to find underlying field

* add a repro

* move querystring-syncing logic to new hook

* rename <Parameters> to <SyncedParametersList>

* Remove syncQueryString prop

* remove unused 'query' prop

* remove 'isQB' prop

* remove ParametersWidget in favor of ParametersList handling empty parameters list

* Rely on value-populated parameters objects over parameterValues map

* remove unused pass-through prop 'vertical'

* make props passed from Dashboard to SyncedParametersList explicit

* Update name of component

* create a more generic hook for syncing the url query string

With the suggestions and highlighting being done through via lexical analysis
only, there is no need to run the computationally-expensive syntactical
analysis on every keystroke anymore. We only need to do that when the
user finishes the editing and wants to use the expression.

* Fix typo

* Don't warn about unused variables prefixed with _

* Fix unused var warning for Icon component

* Add `varsIgnorePattern` for JS files too

* Move the multimethods over to strings not classes

datasets are Cards so don't have a good way to intercept. And this kinda
took all of the keywords, turned them into classes, and then didn't
actually use those classes-- they were just opaque keys in the
hierarchy. No bueno.

* Include datasets in search results

* Docstring for the linter. Thank you clj-kondo

* Fix namespace decls

* Update tests

* Clean up ns and make test more resilient

these tests that test global collections can be sensitive to test order
and the CI dbs that continually exist rather than be
recreated. Sometimes "table" shows up in these available models. Kinda
don't care so just check for subset

* Rename for confusion

* tree view for datasets

* Update docstring

There were global audit query limits which defaulted to 1000 previously. This makes them less limited.

* Expose the e2e downloads helper function

* Apply the new e2e downloads helper to the #18382 repro

* Apply the new e2e downloads helper to the #18440 repro

* Update the helper so it applies both for saved and unsaved questions

* Add additional helper options

* Redirect to a non-existing page

* Reafactor `downloads` spec

* Extract repro for #10803

* Move UI mocks into separate file

* Mock ace and Google Analytics by default

* Add helper to run a test in EE environment

* Add test ID to CacheTTLField

* Migrate tests to new EE helper

Delete Secret instances from the app DB, when the corresponding Database instance is deleted

Updating existing test (and renamed it to `secret-db-details-integration-test`) to confirm this behavior

* fix invalid js

* install typings using typesync

* upgrade classnames to get its own typings

* convert colors, settings, Icon to TS

The possible matching/choices for function names, dimensions, segments,
and metrics are only shown when there is at least ONE matching character.
Thus, when the user hasn't inputted anything yet, there won't be any
suggestions.

* Repro #18630

* Ignore field literals in Join's isValid check

* Don't use Join's `isValid` method at `clean`

* Update the keys we consider to unverify a card

Previously when almost anything changed we would unverify a card. But
this was way too aggresive. We really only want to unverify when a query
changes. Far simpler to specify the keys we do want to consider rather
than the keys to ignore.

Was prompted because when we marked a question as a dataset it would
unverify because the `:display` was changed and also `:dataset`. Neither
of these matter for verification purposes

* Update testing context string [ci skip]

* Empty commit to trigger CI

had skipped ci since it was a docstring change but that prevents merging
sad face :(

`org.apache.sshd` requires `net.i2p.crypto/eddsa` dependency
to work with ED25519 keys.

This worked in 0.40 because `eddsa` was included as a transitive
dependency of another unrelated dependency (`buddy`) but stopped
working in 0.41 because the new version of `buddy` no longer depends
on `eddsa`. Thus, we must explicitly include the dependency.

To prevent this from breaking again, switched one of the test
keys to an ED25519 key.

* move dashboard parameter utils

* Move the ParameterSection type to type/Parameter.js

* Handle long running sync in connection pool hash check

Modify the logic in `db->pooled-connection-spec` to check for the latest `DatabaseInstance` from the app DB in case of a hash mismatch, before deciding that the pool needs to be invalidated (based on changing hash of db details)

This is to handle the case where a long running sync keeps passing in a stale version of the `DatabaseInstance` over a period of many minutes/hours, long after the app DB has received an updated version that was saved through the UI

In this case, the connection pool code will get the latest `DatabaseInstance` from the app DB, on a hash mismatch, then compare that against the in-memory has for the given DB ID.  And if it still doesn't match, then the invalidation kicks in

Updating the test accordingly

* Add dataset icon

* Add getter and setter for Question's `dataset` flag

* Add action to turn a saved question into a dataset

* Add UI for turning a question into a dataset

* Use dataset icon for questions turned into datasets

* Show datasets on collections page

* Change colors for dataset icons in collections

* Test can turn a question into a dataset

* Fix unit test

* Fix color literal

* Hide "Turn into a dataset" button for datasets

* Rerun non table question when turned into a dataset

* This was a mistake

* Make the checkins_interval transient datasets more uniquely named

Append a timestamp suffix (initialized when the namespace was loaded) to transient dataset names

Add verbose comments explaining what we're doing

Add transient-dataset? functions for both BQ drivers to put logic in one place (which currently just see if the dataset is named `checkins_interval_*`

Ditch the "legacydriver_" prefix from the legacy BigQuery driver when creating new datasets; it's no longer needed now that transient datasets will have a much more unique suffix

Update the logic for the transient dataset suffix so that it can be more easily detected/parsed later (for cleanup)

Remove old (more than 2 hours) transient datasets when test extensions namespaces load

Add various docstrings